[
https://issues.apache.org/jira/browse/HBASE-12988?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14542262#comment-14542262
]
Andrew Purtell commented on HBASE-12988:
----------------------------------------
bq. Our WAL format needs to go on a diet. Objects are really fat. Plus we got
pb from the wire from which we make pojos -- WAL* -- and then pb into WAL. Fun.
Rethink?
Rethinking is good.
Let's start from a radical position: we get PB from the wire for requests, so
stuff requests directly into the WAL as received. Cut out the whole PB to pojo
back to PB translations. Would simplify a lot but would be a big rethink too,
since we'd want to encode requests in such a way that they can be concatenated
for group commit, at least. There would be more considerations for request
encoding to be WAL- and replay-friendly. For 2.0 maybe. Would be out of scope
for this issue.
> [Replication]Parallel apply edits on row-level
> ----------------------------------------------
>
> Key: HBASE-12988
> URL: https://issues.apache.org/jira/browse/HBASE-12988
> Project: HBase
> Issue Type: Improvement
> Components: Replication
> Reporter: hongyu bi
> Assignee: hongyu bi
>
> we can apply edits to slave cluster in parallel on table-level to speed up
> replication .
> update : per conversation blow , it's better to apply edits on row-level in
> parallel
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
